forked from php/php-src
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathpcre2_dfa_match.c
4060 lines (3492 loc) · 138 KB
/
pcre2_dfa_match.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2022 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains the external function pcre2_dfa_match(), which is an
alternative matching function that uses a sort of DFA algorithm (not a true
FSM). This is NOT Perl-compatible, but it has advantages in certain
applications. */
/* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved
the performance of his patterns greatly. I could not use it as it stood, as it
was not thread safe, and made assumptions about pattern sizes. Also, it caused
test 7 to loop, and test 9 to crash with a segfault.
The issue is the check for duplicate states, which is done by a simple linear
search up the state list. (Grep for "duplicate" below to find the code.) For
many patterns, there will never be many states active at one time, so a simple
linear search is fine. In patterns that have many active states, it might be a
bottleneck. The suggested code used an indexing scheme to remember which states
had previously been used for each character, and avoided the linear search when
it knew there was no chance of a duplicate. This was implemented when adding
states to the state lists.
I wrote some thread-safe, not-limited code to try something similar at the time
of checking for duplicates (instead of when adding states), using index vectors
on the stack. It did give a 13% improvement with one specially constructed
pattern for certain subject strings, but on other strings and on many of the
simpler patterns in the test suite it did worse. The major problem, I think,
was the extra time to initialize the index. This had to be done for each call
of internal_dfa_match(). (The supplied patch used a static vector, initialized
only once - I suspect this was the cause of the problems with the tests.)
Overall, I concluded that the gains in some cases did not outweigh the losses
in others, so I abandoned this code. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#define NLBLOCK mb /* Block containing newline information */
#define PSSTART start_subject /* Field containing processed string start */
#define PSEND end_subject /* Field containing processed string end */
#include "pcre2_internal.h"
#define PUBLIC_DFA_MATCH_OPTIONS \
(PCRE2_ANCHORED|PCRE2_ENDANCHORED|PCRE2_NOTBOL|PCRE2_NOTEOL|PCRE2_NOTEMPTY| \
PCRE2_NOTEMPTY_ATSTART|PCRE2_NO_UTF_CHECK|PCRE2_PARTIAL_HARD| \
PCRE2_PARTIAL_SOFT|PCRE2_DFA_SHORTEST|PCRE2_DFA_RESTART| \
PCRE2_COPY_MATCHED_SUBJECT)
/*************************************************
* Code parameters and static tables *
*************************************************/
/* These are offsets that are used to turn the OP_TYPESTAR and friends opcodes
into others, under special conditions. A gap of 20 between the blocks should be
enough. The resulting opcodes don't have to be less than 256 because they are
never stored, so we push them well clear of the normal opcodes. */
#define OP_PROP_EXTRA 300
#define OP_EXTUNI_EXTRA 320
#define OP_ANYNL_EXTRA 340
#define OP_HSPACE_EXTRA 360
#define OP_VSPACE_EXTRA 380
/* This table identifies those opcodes that are followed immediately by a
character that is to be tested in some way. This makes it possible to
centralize the loading of these characters. In the case of Type * etc, the
"character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a
small value. Non-zero values in the table are the offsets from the opcode where
the character is to be found. ***NOTE*** If the start of this table is
modified, the three tables that follow must also be modified. */
static const uint8_t coptable[] = {
0, /* End */
0, 0, 0, 0, 0, /* \A, \G, \K, \B, \b */
0, 0, 0, 0, 0, 0, /* \D, \d, \S, \s, \W, \w */
0, 0, 0, /* Any, AllAny, Anybyte */
0, 0, /* \P, \p */
0, 0, 0, 0, 0, /* \R, \H, \h, \V, \v */
0, /* \X */
0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */
1, /* Char */
1, /* Chari */
1, /* not */
1, /* noti */
/* Positive single-char repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* upto, minupto */
1+IMM2_SIZE, /* exact */
1, 1, 1, 1+IMM2_SIZE, /* *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
1+IMM2_SIZE, 1+IMM2_SIZE, /* upto I, minupto I */
1+IMM2_SIZE, /* exact I */
1, 1, 1, 1+IMM2_SIZE, /* *+I, ++I, ?+I, upto+I */
/* Negative single-char repeats - only for chars < 256 */
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* NOT upto, minupto */
1+IMM2_SIZE, /* NOT exact */
1, 1, 1, 1+IMM2_SIZE, /* NOT *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
1+IMM2_SIZE, 1+IMM2_SIZE, /* NOT upto I, minupto I */
1+IMM2_SIZE, /* NOT exact I */
1, 1, 1, 1+IMM2_SIZE, /* NOT *+I, ++I, ?+I, upto+I */
/* Positive type repeats */
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* Type upto, minupto */
1+IMM2_SIZE, /* Type exact */
1, 1, 1, 1+IMM2_SIZE, /* Type *+, ++, ?+, upto+ */
/* Character class & ref repeats */
0, 0, 0, 0, 0, 0, /* *, *?, +, +?, ?, ?? */
0, 0, /* CRRANGE, CRMINRANGE */
0, 0, 0, 0, /* Possessive *+, ++, ?+, CRPOSRANGE */
0, /* CLASS */
0, /* NCLASS */
0, /* XCLASS - variable length */
0, /* REF */
0, /* REFI */
0, /* DNREF */
0, /* DNREFI */
0, /* RECURSE */
0, /* CALLOUT */
0, /* CALLOUT_STR */
0, /* Alt */
0, /* Ket */
0, /* KetRmax */
0, /* KetRmin */
0, /* KetRpos */
0, /* Reverse */
0, /* Assert */
0, /* Assert not */
0, /* Assert behind */
0, /* Assert behind not */
0, /* NA assert */
0, /* NA assert behind */
0, /* ONCE */
0, /* SCRIPT_RUN */
0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
0, 0, /* CREF, DNCREF */
0, 0, /* RREF, DNRREF */
0, 0, /* FALSE, TRUE */
0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
0, 0, /* COMMIT, COMMIT_ARG */
0, 0, 0, /* FAIL, ACCEPT, ASSERT_ACCEPT */
0, 0, 0 /* CLOSE, SKIPZERO, DEFINE */
};
/* This table identifies those opcodes that inspect a character. It is used to
remember the fact that a character could have been inspected when the end of
the subject is reached. ***NOTE*** If the start of this table is modified, the
two tables that follow must also be modified. */
static const uint8_t poptable[] = {
0, /* End */
0, 0, 0, 1, 1, /* \A, \G, \K, \B, \b */
1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */
1, 1, 1, /* Any, AllAny, Anybyte */
1, 1, /* \P, \p */
1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */
1, /* \X */
0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */
1, /* Char */
1, /* Chari */
1, /* not */
1, /* noti */
/* Positive single-char repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1, 1, 1, /* upto, minupto, exact */
1, 1, 1, 1, /* *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
1, 1, 1, /* upto I, minupto I, exact I */
1, 1, 1, 1, /* *+I, ++I, ?+I, upto+I */
/* Negative single-char repeats - only for chars < 256 */
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
1, 1, 1, /* NOT upto, minupto, exact */
1, 1, 1, 1, /* NOT *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
1, 1, 1, /* NOT upto I, minupto I, exact I */
1, 1, 1, 1, /* NOT *+I, ++I, ?+I, upto+I */
/* Positive type repeats */
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
1, 1, 1, /* Type upto, minupto, exact */
1, 1, 1, 1, /* Type *+, ++, ?+, upto+ */
/* Character class & ref repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1, 1, /* CRRANGE, CRMINRANGE */
1, 1, 1, 1, /* Possessive *+, ++, ?+, CRPOSRANGE */
1, /* CLASS */
1, /* NCLASS */
1, /* XCLASS - variable length */
0, /* REF */
0, /* REFI */
0, /* DNREF */
0, /* DNREFI */
0, /* RECURSE */
0, /* CALLOUT */
0, /* CALLOUT_STR */
0, /* Alt */
0, /* Ket */
0, /* KetRmax */
0, /* KetRmin */
0, /* KetRpos */
0, /* Reverse */
0, /* Assert */
0, /* Assert not */
0, /* Assert behind */
0, /* Assert behind not */
0, /* NA assert */
0, /* NA assert behind */
0, /* ONCE */
0, /* SCRIPT_RUN */
0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
0, 0, /* CREF, DNCREF */
0, 0, /* RREF, DNRREF */
0, 0, /* FALSE, TRUE */
0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
0, 0, /* COMMIT, COMMIT_ARG */
0, 0, 0, /* FAIL, ACCEPT, ASSERT_ACCEPT */
0, 0, 0 /* CLOSE, SKIPZERO, DEFINE */
};
/* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,
and \w */
static const uint8_t toptable1[] = {
0, 0, 0, 0, 0, 0,
ctype_digit, ctype_digit,
ctype_space, ctype_space,
ctype_word, ctype_word,
0, 0 /* OP_ANY, OP_ALLANY */
};
static const uint8_t toptable2[] = {
0, 0, 0, 0, 0, 0,
ctype_digit, 0,
ctype_space, 0,
ctype_word, 0,
1, 1 /* OP_ANY, OP_ALLANY */
};
/* Structure for holding data about a particular state, which is in effect the
current data for an active path through the match tree. It must consist
entirely of ints because the working vector we are passed, and which we put
these structures in, is a vector of ints. */
typedef struct stateblock {
int offset; /* Offset to opcode (-ve has meaning) */
int count; /* Count for repeats */
int data; /* Some use extra data */
} stateblock;
#define INTS_PER_STATEBLOCK (int)(sizeof(stateblock)/sizeof(int))
/* Before version 10.32 the recursive calls of internal_dfa_match() were passed
local working space and output vectors that were created on the stack. This has
caused issues for some patterns, especially in small-stack environments such as
Windows. A new scheme is now in use which sets up a vector on the stack, but if
this is too small, heap memory is used, up to the heap_limit. The main
parameters are all numbers of ints because the workspace is a vector of ints.
The size of the starting stack vector, DFA_START_RWS_SIZE, is in bytes, and is
defined in pcre2_internal.h so as to be available to pcre2test when it is
finding the minimum heap requirement for a match. */
#define OVEC_UNIT (sizeof(PCRE2_SIZE)/sizeof(int))
#define RWS_BASE_SIZE (DFA_START_RWS_SIZE/sizeof(int)) /* Stack vector */
#define RWS_RSIZE 1000 /* Work size for recursion */
#define RWS_OVEC_RSIZE (1000*OVEC_UNIT) /* Ovector for recursion */
#define RWS_OVEC_OSIZE (2*OVEC_UNIT) /* Ovector in other cases */
/* This structure is at the start of each workspace block. */
typedef struct RWS_anchor {
struct RWS_anchor *next;
uint32_t size; /* Number of ints */
uint32_t free; /* Number of ints */
} RWS_anchor;
#define RWS_ANCHOR_SIZE (sizeof(RWS_anchor)/sizeof(int))
/*************************************************
* Process a callout *
*************************************************/
/* This function is called to perform a callout.
Arguments:
code current code pointer
offsets points to current capture offsets
current_subject start of current subject match
ptr current position in subject
mb the match block
extracode extra code offset when called from condition
lengthptr where to return the callout length
Returns: the return from the callout
*/
static int
do_callout(PCRE2_SPTR code, PCRE2_SIZE *offsets, PCRE2_SPTR current_subject,
PCRE2_SPTR ptr, dfa_match_block *mb, PCRE2_SIZE extracode,
PCRE2_SIZE *lengthptr)
{
pcre2_callout_block *cb = mb->cb;
*lengthptr = (code[extracode] == OP_CALLOUT)?
(PCRE2_SIZE)PRIV(OP_lengths)[OP_CALLOUT] :
(PCRE2_SIZE)GET(code, 1 + 2*LINK_SIZE + extracode);
if (mb->callout == NULL) return 0; /* No callout provided */
/* Fixed fields in the callout block are set once and for all at the start of
matching. */
cb->offset_vector = offsets;
cb->start_match = (PCRE2_SIZE)(current_subject - mb->start_subject);
cb->current_position = (PCRE2_SIZE)(ptr - mb->start_subject);
cb->pattern_position = GET(code, 1 + extracode);
cb->next_item_length = GET(code, 1 + LINK_SIZE + extracode);
if (code[extracode] == OP_CALLOUT)
{
cb->callout_number = code[1 + 2*LINK_SIZE + extracode];
cb->callout_string_offset = 0;
cb->callout_string = NULL;
cb->callout_string_length = 0;
}
else
{
cb->callout_number = 0;
cb->callout_string_offset = GET(code, 1 + 3*LINK_SIZE + extracode);
cb->callout_string = code + (1 + 4*LINK_SIZE + extracode) + 1;
cb->callout_string_length = *lengthptr - (1 + 4*LINK_SIZE) - 2;
}
return (mb->callout)(cb, mb->callout_data);
}
/*************************************************
* Expand local workspace memory *
*************************************************/
/* This function is called when internal_dfa_match() is about to be called
recursively and there is insufficient working space left in the current
workspace block. If there's an existing next block, use it; otherwise get a new
block unless the heap limit is reached.
Arguments:
rwsptr pointer to block pointer (updated)
ovecsize space needed for an ovector
mb the match block
Returns: 0 rwsptr has been updated
!0 an error code
*/
static int
more_workspace(RWS_anchor **rwsptr, unsigned int ovecsize, dfa_match_block *mb)
{
RWS_anchor *rws = *rwsptr;
RWS_anchor *new;
if (rws->next != NULL)
{
new = rws->next;
}
/* Sizes in the RWS_anchor blocks are in units of sizeof(int), but
mb->heap_limit and mb->heap_used are in kibibytes. Play carefully, to avoid
overflow. */
else
{
uint32_t newsize = (rws->size >= UINT32_MAX/2)? UINT32_MAX/2 : rws->size * 2;
uint32_t newsizeK = newsize/(1024/sizeof(int));
if (newsizeK + mb->heap_used > mb->heap_limit)
newsizeK = (uint32_t)(mb->heap_limit - mb->heap_used);
newsize = newsizeK*(1024/sizeof(int));
if (newsize < RWS_RSIZE + ovecsize + RWS_ANCHOR_SIZE)
return PCRE2_ERROR_HEAPLIMIT;
new = mb->memctl.malloc(newsize*sizeof(int), mb->memctl.memory_data);
if (new == NULL) return PCRE2_ERROR_NOMEMORY;
mb->heap_used += newsizeK;
new->next = NULL;
new->size = newsize;
rws->next = new;
}
new->free = new->size - RWS_ANCHOR_SIZE;
*rwsptr = new;
return 0;
}
/*************************************************
* Match a Regular Expression - DFA engine *
*************************************************/
/* This internal function applies a compiled pattern to a subject string,
starting at a given point, using a DFA engine. This function is called from the
external one, possibly multiple times if the pattern is not anchored. The
function calls itself recursively for some kinds of subpattern.
Arguments:
mb the match_data block with fixed information
this_start_code the opening bracket of this subexpression's code
current_subject where we currently are in the subject string
start_offset start offset in the subject string
offsets vector to contain the matching string offsets
offsetcount size of same
workspace vector of workspace
wscount size of same
rlevel function call recursion level
Returns: > 0 => number of match offset pairs placed in offsets
= 0 => offsets overflowed; longest matches are present
-1 => failed to match
< -1 => some kind of unexpected problem
The following macros are used for adding states to the two state vectors (one
for the current character, one for the following character). */
#define ADD_ACTIVE(x,y) \
if (active_count++ < wscount) \
{ \
next_active_state->offset = (x); \
next_active_state->count = (y); \
next_active_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_ACTIVE_DATA(x,y,z) \
if (active_count++ < wscount) \
{ \
next_active_state->offset = (x); \
next_active_state->count = (y); \
next_active_state->data = (z); \
next_active_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_NEW(x,y) \
if (new_count++ < wscount) \
{ \
next_new_state->offset = (x); \
next_new_state->count = (y); \
next_new_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_NEW_DATA(x,y,z) \
if (new_count++ < wscount) \
{ \
next_new_state->offset = (x); \
next_new_state->count = (y); \
next_new_state->data = (z); \
next_new_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
/* And now, here is the code */
static int
internal_dfa_match(
dfa_match_block *mb,
PCRE2_SPTR this_start_code,
PCRE2_SPTR current_subject,
PCRE2_SIZE start_offset,
PCRE2_SIZE *offsets,
uint32_t offsetcount,
int *workspace,
int wscount,
uint32_t rlevel,
int *RWS)
{
stateblock *active_states, *new_states, *temp_states;
stateblock *next_active_state, *next_new_state;
const uint8_t *ctypes, *lcc, *fcc;
PCRE2_SPTR ptr;
PCRE2_SPTR end_code;
dfa_recursion_info new_recursive;
int active_count, new_count, match_count;
/* Some fields in the mb block are frequently referenced, so we load them into
independent variables in the hope that this will perform better. */
PCRE2_SPTR start_subject = mb->start_subject;
PCRE2_SPTR end_subject = mb->end_subject;
PCRE2_SPTR start_code = mb->start_code;
#ifdef SUPPORT_UNICODE
BOOL utf = (mb->poptions & PCRE2_UTF) != 0;
BOOL utf_or_ucp = utf || (mb->poptions & PCRE2_UCP) != 0;
#else
BOOL utf = FALSE;
#endif
BOOL reset_could_continue = FALSE;
if (mb->match_call_count++ >= mb->match_limit) return PCRE2_ERROR_MATCHLIMIT;
if (rlevel++ > mb->match_limit_depth) return PCRE2_ERROR_DEPTHLIMIT;
offsetcount &= (uint32_t)(-2); /* Round down */
wscount -= 2;
wscount = (wscount - (wscount % (INTS_PER_STATEBLOCK * 2))) /
(2 * INTS_PER_STATEBLOCK);
ctypes = mb->tables + ctypes_offset;
lcc = mb->tables + lcc_offset;
fcc = mb->tables + fcc_offset;
match_count = PCRE2_ERROR_NOMATCH; /* A negative number */
active_states = (stateblock *)(workspace + 2);
next_new_state = new_states = active_states + wscount;
new_count = 0;
/* The first thing in any (sub) pattern is a bracket of some sort. Push all
the alternative states onto the list, and find out where the end is. This
makes is possible to use this function recursively, when we want to stop at a
matching internal ket rather than at the end.
If we are dealing with a backward assertion we have to find out the maximum
amount to move back, and set up each alternative appropriately. */
if (*this_start_code == OP_ASSERTBACK || *this_start_code == OP_ASSERTBACK_NOT)
{
size_t max_back = 0;
size_t gone_back;
end_code = this_start_code;
do
{
size_t back = (size_t)GET(end_code, 2+LINK_SIZE);
if (back > max_back) max_back = back;
end_code += GET(end_code, 1);
}
while (*end_code == OP_ALT);
/* If we can't go back the amount required for the longest lookbehind
pattern, go back as far as we can; some alternatives may still be viable. */
#ifdef SUPPORT_UNICODE
/* In character mode we have to step back character by character */
if (utf)
{
for (gone_back = 0; gone_back < max_back; gone_back++)
{
if (current_subject <= start_subject) break;
current_subject--;
ACROSSCHAR(current_subject > start_subject, current_subject,
current_subject--);
}
}
else
#endif
/* In byte-mode we can do this quickly. */
{
size_t current_offset = (size_t)(current_subject - start_subject);
gone_back = (current_offset < max_back)? current_offset : max_back;
current_subject -= gone_back;
}
/* Save the earliest consulted character */
if (current_subject < mb->start_used_ptr)
mb->start_used_ptr = current_subject;
/* Now we can process the individual branches. There will be an OP_REVERSE at
the start of each branch, except when the length of the branch is zero. */
end_code = this_start_code;
do
{
uint32_t revlen = (end_code[1+LINK_SIZE] == OP_REVERSE)? 1 + LINK_SIZE : 0;
size_t back = (revlen == 0)? 0 : (size_t)GET(end_code, 2+LINK_SIZE);
if (back <= gone_back)
{
int bstate = (int)(end_code - start_code + 1 + LINK_SIZE + revlen);
ADD_NEW_DATA(-bstate, 0, (int)(gone_back - back));
}
end_code += GET(end_code, 1);
}
while (*end_code == OP_ALT);
}
/* This is the code for a "normal" subpattern (not a backward assertion). The
start of a whole pattern is always one of these. If we are at the top level,
we may be asked to restart matching from the same point that we reached for a
previous partial match. We still have to scan through the top-level branches to
find the end state. */
else
{
end_code = this_start_code;
/* Restarting */
if (rlevel == 1 && (mb->moptions & PCRE2_DFA_RESTART) != 0)
{
do { end_code += GET(end_code, 1); } while (*end_code == OP_ALT);
new_count = workspace[1];
if (!workspace[0])
memcpy(new_states, active_states, (size_t)new_count * sizeof(stateblock));
}
/* Not restarting */
else
{
int length = 1 + LINK_SIZE +
((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA ||
*this_start_code == OP_CBRAPOS || *this_start_code == OP_SCBRAPOS)
? IMM2_SIZE:0);
do
{
ADD_NEW((int)(end_code - start_code + length), 0);
end_code += GET(end_code, 1);
length = 1 + LINK_SIZE;
}
while (*end_code == OP_ALT);
}
}
workspace[0] = 0; /* Bit indicating which vector is current */
/* Loop for scanning the subject */
ptr = current_subject;
for (;;)
{
int i, j;
int clen, dlen;
uint32_t c, d;
int forced_fail = 0;
BOOL partial_newline = FALSE;
BOOL could_continue = reset_could_continue;
reset_could_continue = FALSE;
if (ptr > mb->last_used_ptr) mb->last_used_ptr = ptr;
/* Make the new state list into the active state list and empty the
new state list. */
temp_states = active_states;
active_states = new_states;
new_states = temp_states;
active_count = new_count;
new_count = 0;
workspace[0] ^= 1; /* Remember for the restarting feature */
workspace[1] = active_count;
/* Set the pointers for adding new states */
next_active_state = active_states + active_count;
next_new_state = new_states;
/* Load the current character from the subject outside the loop, as many
different states may want to look at it, and we assume that at least one
will. */
if (ptr < end_subject)
{
clen = 1; /* Number of data items in the character */
#ifdef SUPPORT_UNICODE
GETCHARLENTEST(c, ptr, clen);
#else
c = *ptr;
#endif /* SUPPORT_UNICODE */
}
else
{
clen = 0; /* This indicates the end of the subject */
c = NOTACHAR; /* This value should never actually be used */
}
/* Scan up the active states and act on each one. The result of an action
may be to add more states to the currently active list (e.g. on hitting a
parenthesis) or it may be to put states on the new list, for considering
when we move the character pointer on. */
for (i = 0; i < active_count; i++)
{
stateblock *current_state = active_states + i;
BOOL caseless = FALSE;
PCRE2_SPTR code;
uint32_t codevalue;
int state_offset = current_state->offset;
int rrc;
int count;
/* A negative offset is a special case meaning "hold off going to this
(negated) state until the number of characters in the data field have
been skipped". If the could_continue flag was passed over from a previous
state, arrange for it to passed on. */
if (state_offset < 0)
{
if (current_state->data > 0)
{
ADD_NEW_DATA(state_offset, current_state->count,
current_state->data - 1);
if (could_continue) reset_could_continue = TRUE;
continue;
}
else
{
current_state->offset = state_offset = -state_offset;
}
}
/* Check for a duplicate state with the same count, and skip if found.
See the note at the head of this module about the possibility of improving
performance here. */
for (j = 0; j < i; j++)
{
if (active_states[j].offset == state_offset &&
active_states[j].count == current_state->count)
goto NEXT_ACTIVE_STATE;
}
/* The state offset is the offset to the opcode */
code = start_code + state_offset;
codevalue = *code;
/* If this opcode inspects a character, but we are at the end of the
subject, remember the fact for use when testing for a partial match. */
if (clen == 0 && poptable[codevalue] != 0)
could_continue = TRUE;
/* If this opcode is followed by an inline character, load it. It is
tempting to test for the presence of a subject character here, but that
is wrong, because sometimes zero repetitions of the subject are
permitted.
We also use this mechanism for opcodes such as OP_TYPEPLUS that take an
argument that is not a data character - but is always one byte long because
the values are small. We have to take special action to deal with \P, \p,
\H, \h, \V, \v and \X in this case. To keep the other cases fast, convert
these ones to new opcodes. */
if (coptable[codevalue] > 0)
{
dlen = 1;
#ifdef SUPPORT_UNICODE
if (utf) { GETCHARLEN(d, (code + coptable[codevalue]), dlen); } else
#endif /* SUPPORT_UNICODE */
d = code[coptable[codevalue]];
if (codevalue >= OP_TYPESTAR)
{
switch(d)
{
case OP_ANYBYTE: return PCRE2_ERROR_DFA_UITEM;
case OP_NOTPROP:
case OP_PROP: codevalue += OP_PROP_EXTRA; break;
case OP_ANYNL: codevalue += OP_ANYNL_EXTRA; break;
case OP_EXTUNI: codevalue += OP_EXTUNI_EXTRA; break;
case OP_NOT_HSPACE:
case OP_HSPACE: codevalue += OP_HSPACE_EXTRA; break;
case OP_NOT_VSPACE:
case OP_VSPACE: codevalue += OP_VSPACE_EXTRA; break;
default: break;
}
}
}
else
{
dlen = 0; /* Not strictly necessary, but compilers moan */
d = NOTACHAR; /* if these variables are not set. */
}
/* Now process the individual opcodes */
switch (codevalue)
{
/* ========================================================================== */
/* These cases are never obeyed. This is a fudge that causes a compile-
time error if the vectors coptable or poptable, which are indexed by
opcode, are not the correct length. It seems to be the only way to do
such a check at compile time, as the sizeof() operator does not work
in the C preprocessor. */
case OP_TABLE_LENGTH:
case OP_TABLE_LENGTH +
((sizeof(coptable) == OP_TABLE_LENGTH) &&
(sizeof(poptable) == OP_TABLE_LENGTH)):
return 0;
/* ========================================================================== */
/* Reached a closing bracket. If not at the end of the pattern, carry
on with the next opcode. For repeating opcodes, also add the repeat
state. Note that KETRPOS will always be encountered at the end of the
subpattern, because the possessive subpattern repeats are always handled
using recursive calls. Thus, it never adds any new states.
At the end of the (sub)pattern, unless we have an empty string and
PCRE2_NOTEMPTY is set, or PCRE2_NOTEMPTY_ATSTART is set and we are at the
start of the subject, save the match data, shifting up all previous
matches so we always have the longest first. */
case OP_KET:
case OP_KETRMIN:
case OP_KETRMAX:
case OP_KETRPOS:
if (code != end_code)
{
ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0);
if (codevalue != OP_KET)
{
ADD_ACTIVE(state_offset - (int)GET(code, 1), 0);
}
}
else
{
if (ptr > current_subject ||
((mb->moptions & PCRE2_NOTEMPTY) == 0 &&
((mb->moptions & PCRE2_NOTEMPTY_ATSTART) == 0 ||
current_subject > start_subject + mb->start_offset)))
{
if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
else if (match_count > 0 && ++match_count * 2 > (int)offsetcount)
match_count = 0;
count = ((match_count == 0)? (int)offsetcount : match_count * 2) - 2;
if (count > 0) (void)memmove(offsets + 2, offsets,
(size_t)count * sizeof(PCRE2_SIZE));
if (offsetcount >= 2)
{
offsets[0] = (PCRE2_SIZE)(current_subject - start_subject);
offsets[1] = (PCRE2_SIZE)(ptr - start_subject);
}
if ((mb->moptions & PCRE2_DFA_SHORTEST) != 0) return match_count;
}
}
break;
/* ========================================================================== */
/* These opcodes add to the current list of states without looking
at the current character. */
/*-----------------------------------------------------------------*/
case OP_ALT:
do { code += GET(code, 1); } while (*code == OP_ALT);
ADD_ACTIVE((int)(code - start_code), 0);
break;
/*-----------------------------------------------------------------*/
case OP_BRA:
case OP_SBRA:
do
{
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
code += GET(code, 1);
}
while (*code == OP_ALT);
break;
/*-----------------------------------------------------------------*/
case OP_CBRA:
case OP_SCBRA:
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE + IMM2_SIZE), 0);
code += GET(code, 1);
while (*code == OP_ALT)
{
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
code += GET(code, 1);
}
break;
/*-----------------------------------------------------------------*/
case OP_BRAZERO:
case OP_BRAMINZERO:
ADD_ACTIVE(state_offset + 1, 0);
code += 1 + GET(code, 2);
while (*code == OP_ALT) code += GET(code, 1);
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
break;
/*-----------------------------------------------------------------*/
case OP_SKIPZERO:
code += 1 + GET(code, 2);
while (*code == OP_ALT) code += GET(code, 1);
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
break;
/*-----------------------------------------------------------------*/
case OP_CIRC:
if (ptr == start_subject && (mb->moptions & PCRE2_NOTBOL) == 0)
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_CIRCM:
if ((ptr == start_subject && (mb->moptions & PCRE2_NOTBOL) == 0) ||
((ptr != end_subject || (mb->poptions & PCRE2_ALT_CIRCUMFLEX) != 0 )
&& WAS_NEWLINE(ptr)))
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_EOD:
if (ptr >= end_subject)
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
return PCRE2_ERROR_PARTIAL;
else { ADD_ACTIVE(state_offset + 1, 0); }
}
break;
/*-----------------------------------------------------------------*/
case OP_SOD:
if (ptr == start_subject) { ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_SOM:
if (ptr == start_subject + start_offset) { ADD_ACTIVE(state_offset + 1, 0); }
break;
/* ========================================================================== */
/* These opcodes inspect the next subject character, and sometimes
the previous one as well, but do not have an argument. The variable
clen contains the length of the current character and is zero if we are
at the end of the subject. */
/*-----------------------------------------------------------------*/
case OP_ANY:
if (clen > 0 && !IS_NEWLINE(ptr))
{
if (ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])